clmul
tmp = clmul(msgpoly, consts->barrett_reduction_const_1);
return clmul(tmp, consts->barrett_reduction_const_2);
p1 = clmul(m0, consts->fold_across_2_longs_const_hi);
p3 = clmul(m1, consts->fold_across_2_longs_const_lo);